Ahead by 3 with 2 minutes left to play, NE faced a 4th and 1 from their own 49. In a move reminiscent of the infamous 4th and 2 call against the Colts last season, Bill Belichick went for the conversion. RB Ben Jarvis Green-Ellis was dropped two yards shy of the line of scrimmage, turning the ball over on downs to a rallying Chargers team. Was it a good call?
From midfield, punts tend to net 35 yards, giving the Chargers possession at their 16. With just under 2 minutes to go and all 3 timeouts, SD would have a 0.15 WP.
A successful conversion for NE wouldn't seal the win. Because SD still had 3 timeouts, they'd need one more first down. Since it's uncommon that the trailing team would have all 3 timeouts, to estimate NE's win probability properly, we'll dig one level deeper. Teams up by 3 near midfield with 2 minutes to play convert 1st downs 45% of the time. So a successful 4th down conversion seals the game 45% of the time, and it leaves SD with the ball near their own 15 with no timeouts 55% of the time, worth 0.12 WP. This essentially gives SD a 24% chance to get the FG, and then a 50/50 shot in OT. A successful conversion would be worth (to San Diego):
(0.45 * 0) + (0.55 * 0.12) = 0.066 WP
A failed 4th down conversion attempt gives SD the ball at (or near) the NE 49 with about 1:50 and change. That would be worth 0.32 WP for SD.
4th and 1s are converted 74% of the time in that region of the field. That puts the overall WP (for SD) for going for it at:
(0.74 * 0.066) + (0.26 * 0.32) = 0.13 WP
In this case, the baseline numbers say the percentage play would be to go for it, but not by much. The reason it's not a slam-dunk call is because SD had their timeouts. Last year, against the Colts, a conversion would have sealed the win because Indianapolis only had one timeout.
This situation is a great lesson in why going for it is an undervalued strategy. Even if the attempt fails, you still have a very good chance of winning the game. SD did manage to drive into FG range, but missed a 50-yarder.